Who it is for
For anyone who needs more than vendor self-assessments — a practical way to track AI risk and hold every part of their AI stack accountable.
AI Knowledge Series / intermediate
The AI Risk Register
Build a practical AI risk-tracking system that goes beyond vendor self-assessments to establish real accountability across your AI stack.
The problem
It starts with the situation you're actually in.
A procurement form lands in your inbox: a vendor selling your team an AI contract-review tool attaches a security questionnaire, a SOC 2 report, and a two-page "Responsible AI" statement. The gap between what a vendor discloses and what your organization is answerable for is the problem this book solves.
You can fill a shared drive with an inventory, a scored register, an ownership map, a mitigation plan, a framework crosswalk, and a maintenance routine — and still not have a running system. The gap between a finished document and a working discipline is not more documentation.
Outcomes
What you'll be able to do.
Find the AI Tools Nobody Listed
You stop relying on the short, confident list a department head gives you — the licensed chatbot, maybe a coding assistant, maybe the meeting transcriber. You ask the team directly and surface the AI tools actually in use.
Treat Ownership as Three Jobs, Not One
You make sure no risk has a score but no name. When a résumé-screening tool filters out qualified applicants by school or ZIP code, or a pricing tool overcharges a segment, you already know who answers — before leadership asks.
Make Frameworks a View, Not New Work
You answer the auditor who says "Show me how this AI system meets your obligations under the EU AI Act" straight from the register you already keep — no separate scramble.
Spot the Register That Looks Right but Is Wrong
You catch the register that looks healthy from across the room — rows filled, fields complete — but has quietly drifted out of step with reality.
Inside the book
A closer look at the work inside.
Are You Ready to Start the Register?
You are ready to move to another chapter when you can answer yes to most of these for at least one system.
- I named a specific AI system and stopped treating "AI" as one thing.
- I wrote down what the vendor disclosed and when it was last verified.
- I identified at least one risk that exists only because of how we deployed it.
- I can describe how a failure would be detected, or admitted that we could not detect it.
- I named a person who would be accountable, or flagged that no one is.
What This Book Gives You
- Scope mismatch. The vendor assesses the model and the platform. Your risk lives in the integration: the data you feed it, the decisions you let it influence, and the human steps.
- Incentive mismatch. A self-assessment is a sales document. It is written to reduce friction in your buying process, not to surface the failure modes that would make you hesitate.
- Accountability mismatch. When the tool errs, regulators, clients, and your board hold you accountable, not the vendor's responsible-AI page.
Is Your Risk Language Actually Shared?
- Every risk in my map has both a causal read and a domain read, not just a topic label.
- No two categories describe the same harm under different names.
- No single category is being stretched across two genuinely different harms.
- A colleague who did not build the map can read one coordinate line and know where to intervene.
- My harm types map onto recognizable domains rather than ad hoc, project-specific words.
- Each risk could be handed to another team without a glossary attached.
Shared-Language Readiness Scorecard
Use this scorecard after the worksheet. Mark each row weak, usable, or strong, and fix anything weak before moving into the full inventory. For "both taxonomies applied," weak means risks carry only topic labels like "privacy" or "bias"; usable means most risks have causal and domain reads, with a few unclear cases.
Routing One Real Risk Through Both Taxonomies
Take a concrete situation you can verify against your own stack. Your marketing team uses an AI image generator for ad creative. A generated image reproduces a recognizable trademarked logo in the background, and the ad runs for two days before anyone notices.
The AI Risk Register frameworkInside the book
Visual preview
A diagram you can keep open while you work.
Table of contents
9 chapters, built to be read in order.
01
Why Vendor Self-Assessments Aren't Enough
02
Speaking a Common Risk Language
03
Taking Inventory of Your AI Stack
04
Designing the Register's Fields and Scoring
05
Assigning Real Ownership and Accountability
06
Choosing and Applying Mitigations
07
Aligning to Frameworks and Regulations Without Drowning in Them
08
Keeping the Register Alive
09
From Register to Operating Discipline
118
Pages
14,021
Words
56
Exercises, checklists & tools
9
Chapters
Formats
Three formats. One purchase.
EPUB, PDF, and HTML are included so the book can work on an e-reader, as a designed copy, or as a searchable desk reference.
EPUB
For e-readers and reading apps.
The designed edition with diagrams and layouts intact.
HTML
Searchable, copy-pasteable, and practical as a reference.
Complete guide
The AI Risk Register
$9.99
intermediate guide pricing
- ✓EPUB + PDF + HTML included in one purchase
- ✓9 chapters from the complete guide
- ✓DRM-free files for your own devices
- ✓7-day refund review for duplicate purchases, access issues, wrong files, or materially defective downloads
Secure checkout / instant download / tax handled at checkout
Before you buy
Questions, answered.
Does this include the full book?
Yes. You get the complete edition, including the chapter sequence and internal materials described on this page.
Which formats are included?
EPUB, PDF, and HTML are included so you can read on an e-reader, keep a designed copy, or use the searchable browser version.
What is the refund policy?
Because this is an instant digital download, broad change-of-mind refunds are not offered after the files have been accessed. Refund requests are reviewed within 7 days for duplicate purchases, accidental purchases before access, access failures we cannot fix, wrong files, corrupted files, or pages that materially misdescribe the book.